Tuesday June 5th,
14:00
CCC conference room
874/1-011
There were no comments on the minutes of the 26th meeting.
Massimo discussed parameters and tolerance. His talk started by describing the general approach for deriving tolerances, and he then presented several examples for key parameters. He explained that relaxed beam condition during commissioning provide some margin for relaxing tolerances.
In a first step the table of target parameters was reviewed and augmented. For example, transverse IP shifts are needed to maximize aperture. This information is now included in the parameter table. At IP1/5 the IP shift is used for beta*=0.55m, at IP2 and 8 the shift is needed at injection. The values of the crossing angles in IP8 for both spectrometer polarities were also added. A new table for the commissioning stage of 156 bunches was introduced as well. Lastly, the longitudinal emittance at injection into the SPS is 20-30% lower than originally expected, which is now also taken into account.
Roger suggested that the revised and updated target parameter tables be posted on the LHCCWG web site.
Massimo next reviewed the parameter and tolerance table presented by Frank at LHCCWG#19. In the following, he focused on some of the quantities, highlighted by a darker background color.
The design tolerances for the variation of the bunch-to-bunch intensity and emittances have already been achieved in the injector complex. Therefore, these values can be used as input to derive other tolerances. Massimo explained that the situation is more easily assessed for the closed orbit, beta beating and dispersion mismatch. These quantities are all defined and linked via the mechanical aperture. On the other hand, emittance and intensity variation are connected via the luminosity definition.
Jean-Jacques asked for a clarification of the term “tolerances”. Massimo replied that his presentation refers to the relaxation of tolerances with respect to their design specification. The performance of the instrumentation must be sufficient to measure and verify that all tolerances are met.
The required transverse aperture consists of several components, including mechanical tolerances of the beam screen, beta beating, dispersion beating, closed orbit, etc. The relevant definitions were introduced by B. Jeanneret and R. Ostojic in LHC Project Note 111, “Geometrical Acceptance in LHC Version 5.0”. The target is an “n1 value” equal to 7.
Referring to Ralph’s presentation at LHCCWG#5 Massimo pointed out that at injection the primary collimators need to be set at 5.7 sigma. Under nominal conditions no margin is left. During early commissioning the maximum aperture gain is only about 0.5 sigma if all constraints from machine protection are maintained. The only parameter which could be relaxed during the commissioning is the cold aperture, which however must stay above 7 sigma (its nominal value is 7.5 sigma).
The present aperture budget allocates 4 mm of aperture to the closed orbit, 1 mm to beta beating, and another 1 mm to dispersion beating. A larger margin for optics errors can be obtained by assigning the 0.5 sigma (0.6 mm) obtained from a reduced cold aperture to beta beating and dispersion beating only. The tolerances for the latter two could then be increased from 20% to 26%, and from 30% to 39%, respectively. An internal re-distribution of aperture from the closed orbit to the other two sources could raise the beta beating tolerance to 36% and the dispersion beating tolerance to 54%. This represents about a factor two increase in the tolerance .
Brennan commented that in the proposed approach we do not make use of the fact that we can relax the settings of the protection devices. The latter could be set to 10 sigma for 43 bunches. Ralph suggested that these devices can then be taken out completely. Brennan agreed with this comment. Massimo explained that his proposed parameters apply to 43 bunches as well as to 156 bunches. Brennan specified that a margin of 2.5 sigma could be gained. Ralph cautioned that we should be prepared to cope with an arc aperture much smaller than expected. As a safer approach he recommended to decrease the emittance. Brennan underlined that we here discuss the tolerances for the initial commissioning, where even the primary collimator could be retracted. Ralph stressed that we absolutely need to identify the root cause of any problems encountered and not bypass them by compromising the safety of the machine. Brennan recommended addressing problems piecewise. Stephane summarized earlier discussions where the main argument was the need of protection. If protection is no longer an absolute argument, one could relax the tolerances based on cleaning efficiency only. Brennan added that an entire beam would never be kicked since the kick length in this phase is limited to 10 microseconds. Frank suggested that we may have no choice if the actual beta beating turns out to be larger than the tolerance. Ralph S. clarified that the aperture tolerances already include orbit errors, beta beating etc. While collimators are centered on the beam, this is not true for the arcs.
Massimo concluded that every aspect is clear, and that we have all the procedures in hand for computing intensity-dependent tolerances. If we stick to the protection function, the tolerances cannot be changed much for the different phases. Roger suggested that it might be a good approach to start from relaxed tolerances for protection devices with values provided by Brennan.
Ralph reiterated that he preferred lower emittance rather than giving up the protection of the cold aperture. Stephane commented that there would be a 2-sigma space between the TDI and the cold aperture, and only some of the bunches could hit the cold aperture. However, Ralph pointed out that in case of a missing kicker re-triggering the entire beam could be kicked to an amplitude of 10 sigma, which was confirmed by Brennan.
After the meeting, Stephane re-emphasized that reducing the demand on the protection by the TDI does not necessarily mean retracting the TDI. The TDI could stay at its nominal position (7 sigma), even if we allow a cold aperture of less than 7 sigma for small intensity beam.
Now Massimo turned to the issue of dynamic aperture. He explained the origin of a factor two safety which had been used as an LHC design criterion. This safety factor reduces the simulated dynamic aperture of 11-12 sigma to the actually expected dynamic aperture of about 6 sigma. The relation between dynamic aperture and beam lifetime is difficult to predict with certainty, though various models exist. The dependence will need to be measured at the LHC.
The luminosity will deviate from its target value due to fluctuation in bunch intensity, and emittances, and due to beta beating. Massimo assumed that there was no correlation between the two beams, and no correlation between emittance and intensity variations, since the latter two are mainly introduced by the bunch splitting in the PS.
Massimo proposed the tolerance criterion that any additional variation should stay in the shadow of the natural one, e.g. be a factor 2/3 smaller. Summing the various contributions in quadrature would yield a total luminosity variation of about 30%.
Frank commented that another constraint would come from beam-beam, which is known to be very sensitive to unequal emittances or unequal beta functions, even at low values for the beam-beam tune shift. Massimo replied that all tolerances are determined by the performance of the injectors. Roger suggested that at low intensity, with fewer bunch splittings, the injector complex might do better than for nominal beam. Gianlugi remarked that on the contrary he thought that the commissioning beams with lower bunch intensity could be worse in terms of bunch-to-bunch intensity variation.
Next, Massimo returned to the question of beam lifetime. Assuming a luminosity lifetime of 15 h dominated by interactions with the residual gas, a vacuum lifetime of 30-40 h would be acceptable. The residual-gas beam lifetime is inversely proportional to the vacuum pressure. Hence a lifetime decrease by a factor of three with respect to nominal implies a similar increase in the average pressure of the machine.
Another point is the tune control. Based on dynamic aperture tracking studies, the detuning with amplitude should not be relaxed to more than 5e-3 at 6 sigma.
Lastly commenting on the acceptable emittance variation, from the beam-beam tune shift formula and using the nominal and ultimate tune-shift values Massimo derived minimum values for the emittance. Another limit arises from the damage threshold. An upper bound on the emittance is determined by the mechanical aperture. At 450 GeV, the nominal emittance value is close to this limit.
Summing up his tolerance analysis, Massimo concluded that a new iteration of the tolerance table has been completed and a set of extended tolerances established. Criteria for deriving relaxed tolerances were described and example values presented.
Responding to a question by Roger, Massimo confirmed that in the LHC the bunch-to-bunch luminosity variation will never be smaller than about 25%. As stated, before, this estimate is derived from the presently achieved bunch-to-bunch intensity and emittance fluctuations of 10% each. Massimo’s remark was followed by a discussion on the squared addition of emittance variation, intensity variation and beta beating, used for calculating the impact on luminosity. Massimiliano and Witold both commented that the beta beating would not lead to a bunch-to-bunch variation in luminosity.
Stephane later elaborated that, the 25% luminosity variation quoted (22% more exactly) comes from a quadratic addition of the impact of bunch-to-bunch charge and emiitance jitter (10% assumed in both cases) and from the 20% assumed beta*-beat, or in other words, these 25% refer to the possible loss of luminosity w.r.t. the nominal luminosity. But if one wanted to estimate the bunch-to-bunch luminosity variation, the beta-beat contribution shall be removed and a better estimate would be of the order of 15% (at least if the 10% b-to-b charge and emittance jitter are achieved in practice), while the remaining 15-20% (15^2+20^2=25^2) could be the difference of luminosity between ATLAS and CMS...possibly much more problematic!
Frank recalled that at the Tevatron and RHIC the bunch emittances easily change by a factor of two during a fill. Elliott agreed and asked whether a correction of the LHC IP beta function is actually planned.
Ralph asked for the purpose of the tolerance tables. He suggested that we keep our original goal to set up the machine for a safe beam.
Ralph S. suggested that correcting the orbit could be the easiest part of the commissioning. Stephane asked for the reason. Jean-Pierre explained that the design tolerance of 4 mm was derived from the real experience at LEP, adding that at LEP the correction of beta beating and dispersion beating proved much easier than the orbit correction. Ralph S. pointed out that at the LHC the electrical BPM offsets will be smaller than a few 100 microns. Jean-Pierre replied that the LEP experience was consistent with that at many other machines. The quantitative discussion of this topic required a detailed explanation, which Ralph S. sent after the meeting.
=> ACTION: Clarify strategy for setting of collimators and protection devices during commissioning and determine the associated relaxed tolerances (Jan, Brennan, Ralph, Massimo, Jorg, Verena, MPSWG)
Philippe started his talk acknowledging contributions from Magali, Andy and other colleagues. The first part of the presentation addressed the numbering of bunches, reproducing a recent presentation at the LHC Experiment Accelerator Data Exchange Working Group (LEADE) on 7 May. The proposed convention is that bucket no. 1 refers to the first bunch after the abort gap. Proposal number 2 is that for the RF instrumentation to work the first bunch must be in bucket 1.
Jan commented that it might be wise not to place the first bunch near the rising edge of the dump kicker. Frank commented that the first bunch after the abort gap could be untypical and might have a poor lifetime. John concurred, saying that this bunch is a likely candidate to be lost first. Ralph also agreed.
For 25 ns operation, bunches will occupy buckets 1, 11, 21 etc. John suggested that the situation for ions could be different, since they should collide in IP2. For 43 proton bunches, buckets 1, 811, 1621, etc. will be occupied (see “Standard Filling Schemes for Various LHC Operation Modes”, LHC-OP-ES-0003 rev. 1.0).
Stephane clarified that by default, with the proposed
convention, we will always have collisions at IP1. Jean-Jacques suggested that more
flexibility would be desirable and that we might want to inject into other
buckets. Following up a comment from Stephane, Paul
pointed out that this flexibility is
needed right from the start, so that during commissioning we could store beams
in both rings without any bunches encountering each other in the regions with
common vacuum chamber.
Now turning to the revolution frequency and orbit, Philippe explained that at a given place and for constant beam energy the delay between the pulse and the passage of a bunch in bucket 1 will be fixed from run to run. The same statement is true for the bunch clock, which is a square wave obtained via dividing the RF frequency by 10.
It was reiterated that the proposed convention for the collision pattern is that bunches in bucket 1 will collide in IP1 (or, possibly, any other IP, to be agreed upon). Philippe explained that the abort dump receives its own private signal from the RF in point 4, which is always in phase with the abort gap.
The second part of Philippe’s talk addressed the commissioning with beam. A schematic illustrated the low-level RF beam control. Its components include a beam phase module, a beam position module for the radial loop, as well as a synchro-loop which will keep the beams at the center of the vacuum chamber, plus several other modules, in particular a dual frequency program FPGA.
John asked whether the possibility exists to apply radial steering for one of the two beams only, explaining that this would be important for proton-ion collisions. Philippe replied that this is easily possible.
Philippe now stepped through the commissioning phases. Philippe listed the pre-requisites for phase A.1. An injection kicker timing signal will be sent our first. Adjusting the gain of the front end of the beam position and phase modules plus time alignment require 8 hours of dedicated beam time. Next the RF signals are provided for the abort dump and to the experiments.
Massimiliano and Thomas commented that the experiments will start taking beam-gas data from the very first injection.
In A2, the RF cavities will have been phased already without beam, the cavity controller loops will be commissioned, the cavities conditioned to 10 MV/m, and the electronics tested. Goals are centering the closed orbit and getting the pilots to collide at the right point (“cogging”). 8 hours of shift for OP and RF are needed for adjusting the injection frequency. Philippe described details of two alternative procedures to do so. At the end of this phase the first turn is centered and the beam frequency equals the RF frequency. The measurement frequency error is less than 10 Hz. Next the RF is switched on and jumps onto the beam phase-wise. The voltage seen by the beam is not zero in case of large phase errors. The worst case would lead to a loss after 14 ms. For counteracting such losses, now the phase loop will be commissioned, requiring 2 times 8 hours. Frequency and phase errors can be determined in open loop mode allowing corrections of both. Another 8 hours are needed to commission the synchro-loop. The LHC procedure will be similar to the SPS standard procedure, which Philippe illustrated by a couple of scope pictures.
Responding to a question by Roger, Philippe confirmed that these steps are done sequentially for the two beams, and that the 8 hours are the time needed with beam. Parallel commissioning for the two rings is improbable as the SPS conditions (e.g. dipole field) also may need to be changed for the two LHC rings.
Next the beam will be captured, with one cavity at a time over 8 hours, and the closed orbit timing signals are produced. Ralph S. commented that the closed orbit can be measured even without synchronization, by auto-triggering. Two times 4 hours are required for fine tuning of the capture.
Now we will be in a position to align the two rings (4 hours). This stage has been advanced with respect to the original schedule. The collision point is measured and adjusted. There are a number of pick ups which detect signals from both beams. Stephane asked why IP1 is favored, pointing out that the optimum phase for one IP does not need to be the same as the optimum phase in another.
Paul asked whether only the collision timing is controlled, but no attempt is made to observe the collisions directly. Philippe answered that, yes, this is the present assumption. Gianluigi commented that the SPS should be matched to the optimmum integral magnetic field for both LHC rings and for that reason the energy matching SPS-LHC should be performed for both rings at the same time. Massimiliano proposed that one could use a real bunch instead of a pilot bunch to have a larger collision signal. Gianluigi cautioned that the beam lifetime might be extremely low in this phase, of order of a few hundred turns.
The loss of particles from the bucket seen in the RF beam component will be compared with the signal from the fast BCT.
There are two possible approaches to control the radial position, namely via the synchro-loop and via the beam control loops, respectively. Ralph S. asked how many pick ups are used for the radial loop. A single one is foreseen for the start. Philippe commented that an efficient synchrotron phase error compensation using the radial loop scheme would favour momentum correction (frequency shifts) being received at a rate of about 10 Hz. Ralph S. commented that , based on earlier discussions with A. Butterworth, Q. King and J. Wenninger, it is foreseen that the LHC orbit feedback will send these momentum corrections to the RF system as a by-product of the momentum-bias-free orbit correction. While the LHC BPM hardware is capable of acquiring orbit data up to a rate of 25 Hz, Jean-Jacques reminded that the present BI-SW commitment foresees only orbit acquisitions up to 1 Hz during commissioning. 8 hours are needed to commission the radial loop. The radial loop should work before we start the ramp. For multi-bunch injection, the filling pattern mask must be commissioned. Andy explained that only the filling pattern mask will be transmitted to the users, not the entire mask. The pattern is defined via codes. Beam parameters are measured bunch by bunch.
Next, multi-batch injections (8 hours) require bucket number and ring identifier. For a single batch, the phase loop will take care of injection errors. For the second and later batches this is no longer possible, and errors should be taken care of by the longitudinal damper.
Up to 156 bunches and 9e10 / bunch, there
is no need to commission the 1-turn feedback or half detuning.
Ralph asked about the transverse damper. Philippe responded that the latter will be covered in a later LHCCWG presentation by Wolfgang Hofle.
Ramping (3x8 hours for RF) can be done either with the synchro-loop or with the radial loop as a fall back. The cavity Q is changed before the ramp.
Lastly, on the flat top, the voltage is raised to lower the bunch length. Each ring is phased onto the commercial clean 7-TeV synthesizer (2x8 hours). The signal from the synthesizer is also sent to the experiments, already at the moment of injection.
Ralph S. pointed out that the fixed frequency synthesizer could imply significant orbit changes at the collimators, beyond the 35 micron tolerance, due to global circumference changes.
Philippe presented the arguments in favor of a synthesizer. Fine trims can be included by switching between the synthesizer and the dual frequency programme. The switching of references could introduce small perturbations, which can cause longitudinal emittance blow-up. Ralph S. explained that in order to compensate tidal circumference changes, it is necessary to adjust the frequency at least about 10 times within 6 hours to minimize and keep the global dispersion driven orbit changes within 35 um during physics runs.
Stephane asked about measurements of dispersion and chromaticity. Gianluigi and Philippe replied that these could be performed on the “dual frequency program and rephasor FPGA”. Gianluigi commented that with the present system one could ramp up with the radial loop to understand problems in the ramp and resynchronize the two beams before going into collision.
Magali referred to the adjustment of the synchro-loop dynamics, which was listed as not urgent, and she asked when this point should be done at the latest. Philippe replied that this and other non-urgent items could most likely be done parasitically. Otherwise they would be postponed. Gianluigi suggested that adjusting the synchro-loop dynamics may not be possible during phase A2, since the number of turns is likely to be too low.
The proposed bucket tagging conventions were agreed upon. It was however felt important to provide the flexibility of injecting bunches in any bucket outside the abort gap, especially for the commissioning. Also later on, there should be no obligation to have a bunch in bucket no 1, as it might disappear due to bad lifetime. A memo describing the convention will be circulated to the experiments.
In summary, bucket 1 is the first bucket after the abort gap. Bunches in bucket 1 will collide in IP1. This second convention could be changed later.
=> ACTION: Circulate convention memo to experiments (Philippe, Massimiliano?)
Mike’s presentation was cancelled due to lack of time.
Tuesday June 19th, 14:00
CCC conference room 874/1-011
Provisional agenda
Minutes of previous meeting
Matters arising
Update on BI readiness (Jean-Jacques)
Report from MPSC subgroup (Jan)
AOB
Reported by Frank